Originally published as JCO Early Release 10.1200/JCO.2006.09.5943 on August 13 2007

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Timmerman, D. Articles by Bourne, T. Search for Related Content PubMed PubMed Citation Articles by Timmerman, D. Articles by Bourne, T. Related Articles Related Correspondence Social Bookmarking                            What's this?

Inclusion of CA-125 Does Not Improve Mathematical Models Developed to Distinguish Between Benign and Malignant Adnexal Tumors

Dirk Timmerman, Ben Van Calster, Davor Jurkovic, Lil Valentin, Antonia C. Testa, Jean-Pierre Bernard, Caroline Van Holsbeke, Sabine Van Huffel, Ignace Vergote, Tom Bourne

From the Department of Obstetrics and Gynecology, University Hospitals Katholieke Universiteit Leuven; Department of Electrical Engineering, ESAT-SISTA, Katholieke Universiteit Leuven, Belgium; Department of Obstetrics and Gynaecology, King's College Hospital; Department of Obstetrics and Gynaecology, St George's Hospital Medical School, University of London, London, United Kingdom; Department of Obstetrics and Gynaecology, Malmö University Hospital, Malmö, Sweden; Istituto di Clinica Ostetrica e Ginecologica, Università Cattolica del Sacro Cuore, Roma, Italy; and the Service Gyn/Obstétrique, Centre Médical des Pyramides, Maurepas, France

Address reprint requests to Dirk Timmerman, MD, PhD, Department of Obstetrics and Gynecology, University Hospitals, KU Leuven, Herestraat 49, B-3000 Leuven, Belgium; e-mail: dirk.timmerman{at}uz.kuleuven.ac.be

	ABSTRACT

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Purpose To test the value of serum CA-125 measurements alone or as part of a multimodal strategy to distinguish between malignant and benign ovarian tumors before surgery based on a large prospective multicenter study (International Ovarian Tumor Analysis).

Patients and Methods Patients with at least one persistent ovarian mass preoperatively underwent transvaginal ultrasonography using gray scale imaging to assess tumor morphology and color Doppler imaging to obtain indices of blood flow.

Results Data from 809 patients recruited from nine centers were included in the analysis; 567 patients (70%) had benign tumors and 242 (30%) had malignant tumors—of these 152 were primary invasive (62.8%), 52 were borderline malignant (21.5%), and 38 were metastatic (15.7%). A logistic regression model including CA-125 (M2) resulted in an area under the receiver operating characteristic curve (AUC) of 0.934 and did not outperform a published (M1) without serum CA-125 information (AUC, 0.936). Specifically designed new models including CA-125 for premenopausal women (M3) and for postmenopausal women (M4) did not perform significantly better than the model without CA-125 (M1; AUC, 0.891 v AUC, 0.911 and AUC, 0.975 v AUC, 0.949, respectively). In postmenopausal patients, serum CA-125 alone (AUC, 0.920) and the risk of malignancy index (AUC, 0.924) performed very well. Results were very similar when the models were prospectively tested on a group of 345 new patients with adnexal masses of whom 126 had malignant tumors (37%).

Conclusion Adding information on CA-125 to clinical information and ultrasound information does not improve discrimination of mathematical models between benign and malignant adnexal masses.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
The accurate characterization of an ovarian mass before surgery has become increasingly important. Some ovarian tumors can probably be safely left in situ, and most benign tumors can be treated using a minimally invasive approach.^1-3 When the risk of ovarian cancer is perceived to be high, referral to a tertiary oncology center is mandatory in order to optimize the treatment outcome.^4-6

In this report, we focus on the value of using serum CA-125 as a predictor of malignancy in ovarian tumors. The CA-125 antigen is a glycoprotein with a high molecular weight that is expressed by most epithelial ovarian cancers and is recognized by a monoclonal antibody (OC 125). Serum CA125 is raised in 82% of women with ovarian cancer.⁷ As a result, measurements of serum levels of CA-125 have been used as part of the preoperative evaluation of ovarian masses. In 1990, Jacobs et al⁸ combined measurements of serum CA125 with women's menopausal status and ultrasound features of ovarian tumors to estimate the risk of an ovarian tumor being malignant (the risk of malignancy index [RMI]). This test is widely used to triage patients with ovarian tumors for referral to tertiary oncology units.

Several studies have shown that the morphological appearances of an ovarian mass can be used to predict the likelihood of malignancy.^9-11 Mathematical models have been developed to assess the risk of an adnexal tumor being malignant.^12,13 However, when prospectively evaluated these models did not perform very well.^14-17 This may be explained by limitations imposed by the small size of the populations where the models were created, differences in the prevalence of ovarian cancer and other less common adnexal tumors between the populations where the models were created and those where they were tested, and variations in definitions of ultrasound terms and in histological examination of removed tissues. No published mathematical model has outperformed risk estimation made by an experienced sonologist on the basis of subjective evaluation of clinical data and ultrasound findings.^18,19

The aim of this study was to test the value of serum CA-125 measurements alone or as part of mathematical models to distinguish between malignant and benign ovarian tumors before surgery. Therefore the diagnostic performance of models with and without CA-125 as a predicting variable was compared, for the whole data set as well as for pre- and postmenopausal women separately. For this purpose, we use the database of the International Ovarian Tumor Analysis (IOTA) study. The IOTA study is a multicenter study comprising more than 1,000 adnexal tumors examined with ultrasound in a rigidly standardized manner.²⁰

	PATIENTS AND METHODS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
The multicenter IOTA study comprising nine centers has been described in detail before.²⁰ Women with at least one persisting adnexal mass underwent transvaginal gray scale and color Doppler ultrasound examination by an experienced ultrasound examiner using a standardized examination technique and standardized terms and definitions.²¹ Information on more than 50 clinical variables and ultrasound variables was collected in each patient. If a woman had more than one adnexal mass, the mass with the most complicated ultrasound morphology was included in the statistical analysis. Blood samples were drawn for analysis of CA 125, but information on CA 125 values was not necessary for including the patient in the IOTA study. The immunoradiometric assay CA-125 II (Centocor, Malvern, PA; or Cis-Bio, Gif-sur-Yvette, France; or Abbott Axsym System, REF 3B41-22, Abbott Laboratories Diagnostic Division, Abbott Park, IL; or Immuno-l-analyser, Bayer Diagnostics, Tarrytown, NY; or Vidas, bioMérieux, Marcy l'Etoile, France) was used and the results expressed in U/mL. The outcome measure was the histological diagnosis of the surgical specimen of the mass. Malignant ovarian tumors were staged using the criteria recommended by the International Federation of Gynaecology and Obstetrics.²² Exclusion criteria were pregnancy or surgical removal of the mass more than 120 days after the ultrasound examination. Of the 1,066 patients included in the IOTA study, 809 had information on CA 125 and are included in this study.

Statistical Analysis
Statistical analyses were carried out using SAS version 9.1 (SAS Institute Inc, Cary, NC). The data were extensively inspected to detect multicollinearity and outliers. Those patients in the original training set (70% of the 1,066 women, whose data were used to develop a previously published logistic regression model to predict malignancy in adnexal masses [M1]²⁰ with information available on CA125) were used to create new logistic regression models containing CA 125. The original M1 did not retain CA-125 as a variable in the model. New models containing CA-125 were also developed for premenopausal and postmenopausal patients. Automated and manual forward-backward selection methods were used to create logistic regression models to calculate the risk of malignancy in adnexal tumors. A final selection of input variables for a logistic regression model was made using the Akaike information criterion²³ and Bayesian information criterion²³ model fitting indices, five-fold cross validation (CV)²⁴ results on the area under the receiver operating characteristic (ROC) curve (AUC), and the likelihood ratio P value. A model with a good balance for the Akaike information criterion, Bayesian information criterion, and CV indices was selected, and a likelihood ratio of P .05 was preferred for each predictor in the model. Attention was given to the linearity in the logit assumption and to interactions (using a stepwise selection procedure). Interactions were only included if they clearly improved the CV results. The performance of the models developed herein, as well as that of M1, was then assessed in the respective test sets (ie, in those pre- and postmenopausal women in the original test set [30% of 1,066 women] with information available on CA-125). The AUC was used as the main measure of model performance. The models were also evaluated and compared using the partial AUC (ie, the AUC for that part of the ROC curve with specificity more than 75% [pAUC75], and the sensitivity at 75% specificity [Sens75]).²⁵ Ninety-five percent CIs for performance measures as well as 95% CIs of the difference between two models were calculated using the bias-corrected bootstrap method²⁶ based on 500 bootstrap samples of the test set. In addition, the method proposed by DeLong and colleagues²⁷ was used to check for statistically significant differences in AUCs.

The test set AUC, pAUC75, and Sens75 of the model containing CA-125 were compared with those of M1 in all 809 patients, in premenopausal patients only, and in postmenopausal patients only. We decided a priori that if the AUC of the model including CA-125 would not be significantly larger than that of M1, we would conclude that CA-125 did not improve the diagnostic performance in that subgroup. If, however, the model including CA-125 would perform significantly better than M1, a new model without CA-125 would be developed (to replace M1) and the test set AUC of the new models would be compared. The rationale for this strategy is that if the best model including CA-125 does not outperform M1 (which may not be an optimal model for each subgroup, because it is a model created for the whole IOTA population of 1,066 patients) it is not necessary to develop a new model without CA-125 for each group studied.

The new models containing CA-125 as a predicting variable, as well as M1, were then tested on prospectively collected data from a new group of 508 patients with adnexal masses. These had been examined using the same protocol as in the original IOTA study, 287 at the University Hospitals in Leuven (Belgium), 125 at the University Hospital in Malmö (Sweden), and 96 at the Universita Cattolica del Sacro Cuore in Rome (Italy). CA-125 results were available for 345 of these patients, of whom 128 had malignant tumors (37%). The histology of the tumors is presented in Table A1 (online only). The same measures of test performance and the same statistical methods as those described earlier were used.

The original M1²⁰ contained 12 variables: (1) personal history of ovarian cancer (yes = 1, no = 0); (2) current hormone therapy (yes = 1, no = 0); (3) age of the patient in years; (4) maximum diameter of the lesion (mm); (5) the presence of pain during the examination (yes = 1, no = 0); (6) the presence of ascites (yes = 1, no = 0); (7) the presence of blood flow within a solid papillary projection (yes = 1, no = 0); (8) the presence of a purely solid tumor (yes = 1, no = 0); (9) maximal diameter of the solid component (expressed in mm, but with no increase above 50 mm); (10) irregular internal cyst walls (yes = 1, no = 0); (11) the presence of acoustic shadows (yes = 1, no = 0); and (12) the color score (1, 2, 3, or 4). The logistic regression model (M1)²⁰ provided the estimated probability of malignancy for a particular patient with an adnexal tumor. This probability was equal to y = 1/(1+e^-z), where z = –6.7468 + 1.5985 (1) – 0.9983 (2) + 0.0326 (3) + 0.00841 (4) – 0.8577 (5) + 1.5513 (6) + 1.1737 (7) + 0.9281 (8) + 0.0496 (9) + 1.1421 (10) – 2.3550 (11) + 0.4916 (12), and e is the mathematical constant and base value of natural logarithms.

	RESULTS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Of the 809 patients, 567 had benign tumors while 242 had malignant tumors (30%). Appendix Table A1 (online only) provides further information about the histologic diagnoses. Appendix Tables A2, A3, and A4 (online only) show descriptive statistics for the continuous and ordinal variables (number, median, min, max), binary variables (number, %), and categoric variables (number, %), respectively.

Results for the Whole Study Group
Logistic regression models performed better if CA-125 information was used with a log transformation.

The best model including CA-125 as a predictor of malignancy (M2) contained 10 variables: (1) log CA-125; (2) age; (3) maximum diameter of the lesion (bounded at 60 and 170 mm); (4) maximum diameter of the solid component (bounded to the right at 50 mm); (5) color Doppler score; (6) presence of blood flow in papillary projections; (7) presence of irregular cyst walls; (8) personal history of ovarian cancer; (9) presence of ascites; and (10) presence of hormone therapy. The probability of malignancy is estimated as 1/(1+e^-z) where z = –9.1291 + 0.5471 (1) + 0.0287 (2) + 0.0109 (3) + 0.0496 (4) + 0.6490 (5) + 1.323 (6) + 0.9599 (7) + 2.5447 (8) + 0.9589 (9) – 0.8239 (10).

The diagnostic performance of M2, M1, RMI, and CA-125 on the test set (n = 236) and on the prospectively collected data (n = 344) are presented in Tables 1, 2, and 3, and Figure 1. To sum up, M2 including CA-125 as a predicting variable was not superior to M1 that did not include CA-125 as a predicting variable, M2 and M1 were superior to RMI and CA-125, and RMI was superior to CA-125 alone. Further analyses showed that the information on the presence of ascites, the maximum diameter of the lesion, and the maximum diameter of the solid component made values of serum CA-125 redundant.

View larger version (14K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 1. Receiver operating characteristic curves for M1 (model without CA-125), M2 (model with CA-125), the risk of malignancy index (RMI), and CA-125 alone applied to (A) the test population (n = 236) of the International Ovarian Tumor Analysis database and (B) the prospectively collected new patients (n = 345). AUC, area under the curve.

The diagnostic performance of M3, M1, RMI, and CA-125 on the test set (n = 130) and on the prospectively collected data (n = 182) are presented in Tables 1, 2, and 3, and in Figure 2. To sum up, M3 including CA-125 as a predicting variable was not superior to M1 that did not include CA-125 as a predicting variable, both M3 and M1 were superior to RMI and CA-125, and RMI was superior to CA-125 alone.

View larger version (14K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 2. Receiver operating characteristic curves for M1 (model without CA-125), M3 (model with CA-125), the risk of malignancy index (RMI), and CA-125 alone applied to (A) the test population of premenopausal women (n = 130) of the International Ovarian Tumor Analysis database and (B) the prospectively collected new patients (n = 183). AUC, area under the curve.

The diagnostic performance of M4, M1, RMI, and CA-125 on the test set (n = 106) and on the prospectively collected data (n = 162) are presented in Tables 1, 2, and 3, and in Figure 3. To sum up, the performance of M4 including CA-125 as a predicting variable was similar to that of M1 that did not include CA-125 as a predicting variable. Even though all models performed very well, M4 was superior to RMI and CA-125 in both data sets. The same was true of M1 but differences in performance were smaller, especially the difference between M1 and RMI. RMI and CA-125 had very similar performance. Because there was a difference of 2.6% in AUC of M4 and M1 in the test set, we did develop a new model without CA-125 and compared its performance to that of M4. The difference in test set AUC was 2.4% (0.951 v 0.975; 95% CI for the difference in AUC, – 0.054 to 0.009; P = .13) whereas the difference in prospective AUC was 2.3% (0.936 v 0.959; 95% CI, –0.050 to 0.007; P = .12).

View larger version (13K): [in this window] [in a new window] [PowerPoint Slide for Teaching]   Fig 3. Receiver operating characteristic curves for M1 (model without CA-125), M4 (model with CA-125), the risk of malignancy index (RMI), and CA-125 alone applied to (A) the test population of postmenopausal women (n = 106) of the International Ovarian Tumor Analysis database and (B) the prospectively collected new patients (n = 162). AUC, area under the curve.

	DISCUSSION

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

In premenopausal patients, serum CA-125 alone has very little value (AUC, 0.632) and also the RMI is rather weak (AUC, 0.778),whereas a logistic regression model without CA-125 (M1) has a very good capacity to distinguish between benign and malignant masses (AUC, 0.911). The situation is different in postmenopausal patients where serum CA-125 alone performs very well (AUC, 0.920) and where RMI is equally reliable (AUC, 0.924). But even in postmenopausal patients a logistic regression model M1 without CA-125 information (AUC, 0.949) did not perform much worse than a new model M4 with CA-125 included (AUC, 0.975). On prospective testing in a new set of postmenopausal patients, logistic regression model M1 and M4 had similar AUCs (0.945 and 0.959, respectively), but M1 without CA 125 had a significantly higher sensitivity for malignancy (97.7%) at a 75% specificity level than M4, RMI, and CA-125 alone (93.0%, 89.5%, and 90.7%, respectively). The difference in performance of CA-125 alone and RMI between pre- and postmenopausal patients is in agreement with previous reports^36-40 and may be explained by the different mix of ovarian tumors found in pre- and postmenopausal patients.

We acknowledge that CA-125 was not measured in all 1,066 patients of the IOTA study, and that therefore our study population of 809 patients is a selected sample from the IOTA database. However, an analysis of the original IOTA database of 1,066 patients shows that CA-125 was often not measured when the ultrasound examiner was certain about whether the mass was benign or malignant. In cases where CA-125 was not measured, subjective assessment with regard to benignity versus malignancy performed very well. We do not believe that excluding cases where the diagnosis was so clear that the clinician found it unnecessary to measure CA-125 invalidates our results. Our study population is likely to be representative of patients in whom clinicians consider CA-125 to be helpful. We also recognize that there would probably have been less variation in CA-125 values had CA-125 been assayed using the same kit in every single case, but we believe that our use of different kits reflects clinical practice and is unlikely to have substantially biased our results.

There are several reasons that might explain why we have found CA-125 to be of little value in our study population. Previous models have always used a relatively small number of input variables. In the IOTA study, 55 variables were tested and 12 significant variables were retained in the model. We were able to do this because of the large number of patients in the study. To the best of our knowledge, this is the first study where clinical variables such as a personal history of ovarian cancer, current hormone therapy, and the presence of pain during the ultrasound scan were included in the model building. However, it seems that in our models the information on the presence of ascites, the maximum diameter of the lesion, and the maximum diameter of the solid component makes values of serum CA-125 redundant. If these three variables are removed from the model, CA-125 becomes an independently significant variable. This demonstrates that ultrasound information is more informative than CA-125 when evaluating the risk of an adnexal mass being benign or malignant. Associations between serum CA-125 and both ascites and tumor size have been reported in other studies.^41,42

Finally, our data raise interesting questions about current clinical practice and protocols used to assess ovarian and other adnexal pathology. It is well proven that in experienced hands the subjective impression of an ovarian mass using ultrasonography is very accurate.^18,19 However, not all ultrasound practitioners will perform at this level. Therefore, research into predictive algorithms that help the less experienced examiner is very important. The RMI has been validated in several prospective studies and overall it has demonstrated a reasonable test performance.^14,16,43 However, there is room for improvement. We believe the IOTA data suggest that in the presence of a competent ultrasound unit and with knowledge of certain clinical information about the patient, knowledge of the serum CA-125 level is unlikely to improve the preoperative characterization of a mass or change management, especially in premenopausal patients. In premenopausal patients, measurements of serum CA-125 seem to be useless to distinguish between benign and malignant ovarian tumors and mathematical models without CA-125 could be used. In postmenopausal patients, one could use CA-125 alone, the RMI, or mathematical models. However, even in postmenopausal patients, a mathematical model without CA-125 has the highest sensitivity for malignancy at the 75% specificity level.

	AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
The author(s) indicated no potential conflicts of interest.

	AUTHOR CONTRIBUTIONS

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
Conception and design: Dirk Timmerman, Lil Valentin, Antonia C. Testa, Sabine Van Huffel, Tom Bourne

Financial support: Dirk Timmerman

Provision of study materials or patients: Dirk Timmerman, Davor Jurkovic, Lil Valentin, Antonia C. Testa, Jean-Pierre Bernard, Caroline Van Holsbeke, Ignace Vergote

Collection and assembly of data: Dirk Timmerman, Davor Jurkovic, Lil Valentin, Antonia C. Testa, Jean-Pierre Bernard, Caroline Van Holsbeke

Data analysis and interpretation: Dirk Timmerman, Ben Van Calster, Lil Valentin, Caroline Van Holsbeke, Sabine Van Huffel, Tom Bourne

Manuscript writing: Dirk Timmerman, Ben Van Calster, Lil Valentin, Tom Bourne

Final approval of manuscript: Dirk Timmerman, Ben Van Calster, Davor Jurkovic, Lil Valentin, Antonia C. Testa, Jean-Pierre Bernard, Caroline Van Holsbeke, Sabine Van Huffel, Ignace Vergote, Tom Bourne

Other: Dirk Timmerman [Coordinator of the International Ovarian Tumor Analysis study]

	Appendix

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
International Ovarian Tumor Analysis steering committee. Dirk Timmerman, Lil Valentin, Tom Bourne, Antonia Testa, Sabine Van Huffel, Ignace Vergote; IOTA principal investigators (alphabetical order): Jean-Pierre Bernard, Maurepas, France, Tom Bourne, London, United Kingdom, Enrico Ferrazzi, Milan, Italy, Davor Jurkovic, London, United Kingdom, Fabrice Lécuru, Paris, France, Andrea Lissoni, Monza, Italy, Ulrike Metzger, Paris, France, Dario Paladini, Napels, Italy, Antonia Testa, Roma, Italy, Dirk Timmerman, Leuven, Belgium, Lil Valentin, Malmö, Sweden, Sabine Van Huffel, Leuven, Belgium, Caroline Van Holsbeke, Leuven, Belgium, Ignace Vergote, Leuven, Belgium and Gerardo Zanetta (deceased), Monza, Italy.

	NOTES

 
published online ahead of print at www.jco.org on August 13, 2007.

Supported by GOA-AMBioRICS, CoE EF/05/006, Belgian network DYSCO, BIOPATTERN (FP6-2002-IST 508803), ETUMOUR (FP6-2002-LIFESCIHEALTH 503094), by the Swedish Medical Research Council (Grants No. K98-17X-11605-03A, K2001-72X-11605-06A and K2002-72X-11605-07B), by Malmö University Hospital, the Malmö General Hospital Foundation for Fighting Against Cancer, and ALF-medel.

Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.

	REFERENCES

TOP ABSTRACT INTRODUCTION PATIENTS AND METHODS RESULTS DISCUSSION AUTHORS' DISCLOSURES OF... AUTHOR CONTRIBUTIONS Appendix REFERENCES

 
1. Buchweitz O, Matthias S, Muller-Steinhardt M, et al: Laparoscopy in patients over 60 years old: A prospective randomized evaluation of laparoscopic versus open adnexectomy. Am J Obstet Gynecol 193:1364-1368, 2005[CrossRef][Medline]

2. Medeiros LR, Fachel JM, Garry R, et al: Laparoscopy versus laparotomy for benign ovarian tumours. Cochrane Database Syst Rev 20:CD004751, 2005

3. Carley ME, Klingele CJ, Gebhart JB, et al: Laparoscopy versus laparotomy in the management of benign unilateral adnexall masses. J Am Assoc Gynecol Laparosc 9:321-326, 2002[CrossRef][Medline]

4. Hacker NF, Berek JS, Lagasse LD: Primary cytoreductive surgery for epithelial ovarian cancer. Obstet Gynecol 61:431-420, 1983

5. Sutton GP, Stehman FB, Einhorn LH: Ten-year follow-up of patients receiving cisplatin, doxorubicin, and cyclophosphamide chemotherapy for advanced epithelial ovarian carcinoma. J Clin Oncol 7:223-229, 1989[Abstract]

6. Redman JR, Petroni GR, Saigo PE: Prognostic factors in advanced ovarian carcinoma. J Clin Oncol 4:515-523, 1986[Abstract/Free Full Text]

7. Bast R, Klug T, St John E, et al: A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 309:883-887, 1983[Abstract]

8. Jacobs I, Oram D, Fairbanks J, et al: A risk of malignancy index incorporating CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol 97:922-929, 1990[Medline]

9. Granberg S, Wikland M, Jansson I: Macroscopic characterization of ovarian tumors and the relation to the histological diagnosis: Criteria to be used for ultrasound evaluation. Gynecol Oncol 35:139-144, 1989[CrossRef][Medline]

10. Lerner JP, Timor-Tritsch IE, Federman A, et al: Transvaginal ultrasonographic characterization of ovarian masses with an improved, weighted scoring system. Am J Obstet Gynecol 170:81-85, 1994[Medline]

11. Ferrazzi E, Zanetta G, Dordoni D, et al: Transvaginal ultrasonographic chracterirzation of ovarian masses: Comparison of five scoring systems in a multicenter study. Ultrasound Obstet Gynecol 10:192-197, 1997[CrossRef][Medline]

12. Timmerman D, Verrelst H, Bourne TH, et al: Artificial neural network models for the pre-operative discrimination between malignant and benign adnexal masses. Ultrasound Obstet Gynecol 13:19-25, 1999

13. Tailor A, Jurkovic D, Bourne TH, et al: Sonographic prediction of malignancy in adnexal masses using an artificial neural network. Br J Obstet Gynaecol 106:21-30, 1999[Medline]

14. Timmerman D: Ultrasonography in the assessment of ovarian and tamoxifen-associated endometrial pathology. Thesis (ISBN 906186 857 2); Leuven, Belgium, Leuven University Press, 1997

15. Aslam N, Banerjee S, Carr JV, et al: Prospective evaluation of logistic regression models for the diagnosis of ovarian cancer. Obstet Gynecol 96:75-80, 2000[CrossRef][Medline]

16. Mol BWJ, Boll D, De Kanter M, et al: Distinguishing the benign and malignant adnexal mass: An external validation of prognostic models. Gynecol Oncol 80:162-167, 2001[CrossRef][Medline]

17. Valentin L, Hagen B, Tingulstad S, et al: Comparison of ‘pattern recognition’ and logistic regression models for discrimination between benign and malignant pelvic masses: A prospective cross validation. Ultrasound Obstet Gynecol 18:357-365, 2001[CrossRef][Medline]

18. Timmerman D, Schwärzler P, Collins WP, et al: Subjective assessment of adnexal masses using ultrasonography: An analysis of interobserver variability and experience. Ultrasound Obstet Gynecol 13:11-16, 1999[CrossRef][Medline]

19. Valentin L: Pattern recognition of pelvic masses by gray-scale ultrasound imaging: The contribution of Doppler ultrasound. Ultrasound Obstet Gynecol 14:338-347, 1999[CrossRef][Medline]

20. Timmerman D, Testa AC, Bourne T, et al: Logistic regression model to distinguish between the benign and malignant adnexal mass before surgery: A multicenter study by the International Ovarian Tumor Analysis Group. J Clin Oncol 23:8794-8801, 2005[Abstract/Free Full Text]

21. Timmerman D, Valentin L, Bourne TH, et al: Terms, definitions and measurements to describe the ultrasonographic features of adnexal tumors: A consensus opinion from the international ovarian tumor analysis (IOTA) group. Ultrasound Obstet Gynecol 16:500-505, 2000[CrossRef][Medline]

22. Heintz APM, Odicino F, Maisonneuve P, et al: Carcinoma of the Ovary, in 25th Annual Report on the Results of Treatment in Gynecological Cancer. Int J Gynecol Obstet 83:S135-S137, 2003 (suppl 1)

23. Burnham KP, Anderson DR: Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach (ed 2). New York, NY, Springer-Verlag, 2002

24. Stone M: Asymptotics for and against cross-validation. Biometrika 64:29-35, 1977[Abstract/Free Full Text]

25. Pepe MS, Longton G, Anderson GL, et al: Selecting differentially expressed genes from microarray experiments. Biometrics 59:133-142, 2003[CrossRef][Medline]

26. Chernick ME: Bootstrap methods: A practitioner's guide. New York, NY, Wiley, 1999

27. DeLong ER, DeLong DM, Clarke-Pearson DL: Comparing the areas under two or more correlated receiver operating characteristics curves: A nonparametric approach. Biometrics 44:837-845, 1988[CrossRef][Medline]

28. Clinical Standards Board for Scotland: Clinical Standards for Gynaecological (Ovarian) Cancer 2001. http://www.nhshealthquality.org/nhsqis/files/Ovarian_Cancer%20jan%2001.pdf

29. Department of Health. Guidance on Commissioning Cancer Services: Improving Outcomes in Gynaecological Cancer–The Manual. London, United Kingdom, Department of Health, 1999

30. Buys SS, Partridge F, Greene MH, et al: Ovarian cancer screening in the Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening trial: Findings from the initial screen of a randomized trial. Am J Obstet Gynecol 193:1630-1639, 2005[CrossRef][Medline]

31. Bailey J, Tailor A, Naik R, et al: Risk of malignancy index for referral of ovarian cancer cases to a tertiary center: Does it identify the correct cases? Int J Gynecol Cancer 16:30-34, 2006[CrossRef][Medline]

32. Chan DW, Shih LM, Sokoll L, et al: National Academy of clinical biochemistry guidelines for the use of tumor markers in ovarian cancer. http://www.nacb.org/lmpg/tumor/chp3e_ovarian.pdf

33. National Comprehensive Cancer Network: Ovarian cancer: NCCN clinical practice guidelines in oncology. V. 1.2006. http://www.nccn.org/professionals/physician_gls/PDF/ovarian.pdf

34. Tailor A, Jurkovic D, Bourne TH, et al: Sonographic prediction of malignancy in adnexal masses using multivariate logistic regression analysis. Ultrasound Obstet Gynecol 10:41-47, 1997[CrossRef][Medline]

35. Timmerman D, Bourne T, Tailor A, et al: A comparison of methods for the pre-operative discrimination between benign and malignant adnexal masses: The development of a new logistic regression model. Am J Obstet Gynecol 181:57-65, 1999[CrossRef][Medline]

36. Antonic J, Rakar S: Validity of colour and pulsed Doppler US and tumour marker CA 125 in differentiation between benign and malignant ovarian masses. Eur J Gynaecol Oncol 17:29-35, 1996[Medline]

37. Strigini F, Gadducci A, Del Bravo B, et al: Differential diagnosis of adnexal masses with transvaginal sonography, color flow imaging and serum CA 125 assay in pre-and postmenopausal women. Gynaecologic Oncology 61:68-72, 1996[CrossRef][Medline]

38. Malkasian G, Knapp R, Lavin P, et al: Preoperative evaluation of serum CA125 levels in premenopausal and postmenopausal patients with pelvic masses: Discrimination of benign from malignant disease. Am J Obstet Gynecol 159:341-346, 1988[Medline]

39. Eltabbakh G, Belinson J, Kennedy A, et al: Serum CA-125 measurements >65 U/ml, clinical value. J Reprod Med 42:617-624, 1997[Medline]

40. Vasilev S, Schlaerth J, Campeau J, et al: Serum CA 125 levels in preoperative evaluation of pelvic masses. Obstet Gynecol 71:751-756, 1988[Medline]

41. Tuxen MK: Tumor marker CA 125 in ovarian cancer. J Tumor Markers Oncol 16:49-68, 2001

42. Cooper B, Sood A, Davis C, et al: Preoperative CA 125 levels: An independent prognostic factor for epithelial ovarian cancer. Obstet Gynecol 100:59-64, 2002[CrossRef][Medline]

43. Timmerman D: The use of mathematical models to evaluate pelvic masses: Can they beat an expert operator? Best Pract Res Clin Obstet Gynaecol 18:91-104, 2004[CrossRef][Medline]

Submitted October 20, 2006; accepted April 4, 2007.

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

Related Correspondence

• Pitfalls of CA-125 Levels in the Preoperative Work-Up of Ovarian Masses
  Denis Querleu, Eliane Mery, Gwenael Ferron, Virginia Benito, Arash Rafii, and Laurence Gladieff
  JCO 2008 26: 512 [Full Text]

This article has been cited by other articles:

				D. Querleu, E. Mery, G. Ferron, V. Benito, A. Rafii, and L. Gladieff Pitfalls of CA-125 Levels in the Preoperative Work-Up of Ovarian Masses J. Clin. Oncol., January 20, 2008; 26(3): 512 - 512. [Full Text] [PDF]

				B. Van Calster, D. Timmerman, T. Bourne, A. C. Testa, C. Van Holsbeke, E. Domali, D. Jurkovic, P. Neven, S. Van Huffel, and L. Valentin Discrimination Between Benign and Malignant Adnexal Masses by Specialist Ultrasound Examination Versus Serum CA-125 J Natl Cancer Inst, November 21, 2007; 99(22): 1706 - 1714. [Abstract] [Full Text] [PDF]

				R. G. Moore and R. C. Bast Jr How Do You Distinguish a Malignant Pelvic Mass From a Benign Pelvic Mass? Imaging, Biomarkers, or None of the Above J. Clin. Oncol., September 20, 2007; 25(27): 4159 - 4161. [Full Text] [PDF]

This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Save to my personal folders Download to citation manager Rights & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Timmerman, D. Articles by Bourne, T. Search for Related Content PubMed PubMed Citation Articles by Timmerman, D. Articles by Bourne, T. Related Articles Related Correspondence Social Bookmarking                            What's this?